| | | 1 | | // Copyright (c) ZeroC, Inc. |
| | | 2 | | |
| | | 3 | | using System.Diagnostics; |
| | | 4 | | using System.Globalization; |
| | | 5 | | |
| | | 6 | | namespace Ice.UtilInternal; |
| | | 7 | | |
| | | 8 | | public sealed class Options |
| | | 9 | | { |
| | | 10 | | private enum State |
| | | 11 | | { |
| | | 12 | | Normal, |
| | | 13 | | DoubleQuote, |
| | | 14 | | SingleQuote, |
| | | 15 | | ANSIQuote |
| | | 16 | | } |
| | | 17 | | |
| | | 18 | | public static string[] |
| | | 19 | | split(string line) |
| | | 20 | | { |
| | 1 | 21 | | string IFS = " \t\n"; |
| | | 22 | | |
| | 1 | 23 | | string l = line.Trim(); |
| | 1 | 24 | | if (l.Length == 0) |
| | | 25 | | { |
| | 1 | 26 | | return []; |
| | | 27 | | } |
| | | 28 | | |
| | 1 | 29 | | State state = State.Normal; |
| | | 30 | | |
| | 1 | 31 | | string arg = ""; |
| | 1 | 32 | | var vec = new List<string>(); |
| | 1 | 33 | | for (int i = 0; i < l.Length; ++i) |
| | | 34 | | { |
| | 1 | 35 | | char c = l[i]; |
| | | 36 | | switch (state) |
| | | 37 | | { |
| | | 38 | | case State.Normal: |
| | | 39 | | { |
| | | 40 | | switch (c) |
| | | 41 | | { |
| | | 42 | | case '\\': |
| | | 43 | | { |
| | | 44 | | // |
| | | 45 | | // Ignore a backslash at the end of the string, |
| | | 46 | | // and strip backslash-newline pairs. If a |
| | | 47 | | // backslash is followed by a space, single quote, |
| | | 48 | | // double quote, or dollar sign, we drop the backslash |
| | | 49 | | // and write the space, single quote, double quote, |
| | | 50 | | // or dollar sign. This is necessary to allow quotes |
| | | 51 | | // to be escaped. Dropping the backslash preceding a |
| | | 52 | | // space deviates from bash quoting rules, but is |
| | | 53 | | // necessary so we don't drop backslashes from Windows |
| | | 54 | | // path names.) |
| | | 55 | | // |
| | 1 | 56 | | if (i < l.Length - 1 && l[++i] != '\n') |
| | | 57 | | { |
| | 1 | 58 | | switch (l[i]) |
| | | 59 | | { |
| | | 60 | | case ' ': |
| | | 61 | | case '$': |
| | | 62 | | case '\'': |
| | | 63 | | case '"': |
| | | 64 | | { |
| | 1 | 65 | | arg += l[i]; |
| | 1 | 66 | | break; |
| | | 67 | | } |
| | | 68 | | default: |
| | | 69 | | { |
| | 0 | 70 | | arg += '\\'; |
| | 0 | 71 | | arg += l[i]; |
| | 0 | 72 | | break; |
| | | 73 | | } |
| | | 74 | | } |
| | | 75 | | } |
| | | 76 | | break; |
| | | 77 | | } |
| | | 78 | | case '\'': |
| | | 79 | | { |
| | 1 | 80 | | state = State.SingleQuote; |
| | 1 | 81 | | break; |
| | | 82 | | } |
| | | 83 | | case '"': |
| | | 84 | | { |
| | 1 | 85 | | state = State.DoubleQuote; |
| | 1 | 86 | | break; |
| | | 87 | | } |
| | | 88 | | case '$': |
| | | 89 | | { |
| | 1 | 90 | | if (i < l.Length - 1 && l[i + 1] == '\'') |
| | | 91 | | { |
| | 1 | 92 | | state = State.ANSIQuote; // Bash uses $'<text>' to allow ANSI escape sequences |
| | | 93 | | // within <text>. |
| | 1 | 94 | | ++i; |
| | | 95 | | } |
| | | 96 | | else |
| | | 97 | | { |
| | 0 | 98 | | arg += '$'; |
| | | 99 | | } |
| | 0 | 100 | | break; |
| | | 101 | | } |
| | | 102 | | default: |
| | | 103 | | { |
| | 1 | 104 | | if (IFS.Contains(l[i], StringComparison.Ordinal)) |
| | | 105 | | { |
| | 1 | 106 | | vec.Add(arg); |
| | 1 | 107 | | arg = ""; |
| | | 108 | | |
| | | 109 | | // |
| | | 110 | | // Move to start of next argument. |
| | | 111 | | // |
| | 1 | 112 | | while (++i < l.Length && IFS.Contains(l[i], StringComparison.Ordinal)) |
| | | 113 | | { |
| | | 114 | | } |
| | 1 | 115 | | --i; |
| | | 116 | | } |
| | | 117 | | else |
| | | 118 | | { |
| | 1 | 119 | | arg += l[i]; |
| | | 120 | | } |
| | 1 | 121 | | break; |
| | | 122 | | } |
| | | 123 | | } |
| | | 124 | | break; |
| | | 125 | | } |
| | | 126 | | case State.DoubleQuote: |
| | | 127 | | { |
| | | 128 | | // |
| | | 129 | | // Within double quotes, only backslash retains its special |
| | | 130 | | // meaning, and only if followed by double quote, backslash, |
| | | 131 | | // or newline. If not followed by one of these characters, |
| | | 132 | | // both the backslash and the character are preserved. |
| | | 133 | | // |
| | 1 | 134 | | if (c == '\\' && i < l.Length - 1) |
| | | 135 | | { |
| | 1 | 136 | | switch (c = l[++i]) |
| | | 137 | | { |
| | | 138 | | case '"': |
| | | 139 | | case '\\': |
| | | 140 | | case '\n': |
| | | 141 | | { |
| | 1 | 142 | | arg += c; |
| | 1 | 143 | | break; |
| | | 144 | | } |
| | | 145 | | default: |
| | | 146 | | { |
| | 1 | 147 | | arg += '\\'; |
| | 1 | 148 | | arg += c; |
| | 1 | 149 | | break; |
| | | 150 | | } |
| | | 151 | | } |
| | | 152 | | } |
| | 1 | 153 | | else if (c == '"') // End of double-quote mode. |
| | | 154 | | { |
| | 1 | 155 | | state = State.Normal; |
| | | 156 | | } |
| | | 157 | | else |
| | | 158 | | { |
| | 1 | 159 | | arg += c; // Everything else is taken literally. |
| | | 160 | | } |
| | 1 | 161 | | break; |
| | | 162 | | } |
| | | 163 | | case State.SingleQuote: |
| | | 164 | | { |
| | 1 | 165 | | if (c == '\'') // End of single-quote mode. |
| | | 166 | | { |
| | 1 | 167 | | state = State.Normal; |
| | | 168 | | } |
| | | 169 | | else |
| | | 170 | | { |
| | 1 | 171 | | arg += c; // Everything else is taken literally. |
| | | 172 | | } |
| | 1 | 173 | | break; |
| | | 174 | | } |
| | | 175 | | case State.ANSIQuote: |
| | | 176 | | { |
| | | 177 | | switch (c) |
| | | 178 | | { |
| | | 179 | | case '\\': |
| | | 180 | | { |
| | 1 | 181 | | if (i == l.Length - 1) |
| | | 182 | | { |
| | | 183 | | break; |
| | | 184 | | } |
| | 1 | 185 | | switch (c = l[++i]) |
| | | 186 | | { |
| | | 187 | | // |
| | | 188 | | // Single-letter escape sequences. |
| | | 189 | | // |
| | | 190 | | case 'a': |
| | | 191 | | { |
| | 0 | 192 | | arg += '\a'; |
| | 0 | 193 | | break; |
| | | 194 | | } |
| | | 195 | | case 'b': |
| | | 196 | | { |
| | 0 | 197 | | arg += '\b'; |
| | 0 | 198 | | break; |
| | | 199 | | } |
| | | 200 | | case 'f': |
| | | 201 | | { |
| | 0 | 202 | | arg += '\f'; |
| | 0 | 203 | | break; |
| | | 204 | | } |
| | | 205 | | case 'n': |
| | | 206 | | { |
| | 0 | 207 | | arg += '\n'; |
| | 0 | 208 | | break; |
| | | 209 | | } |
| | | 210 | | case 'r': |
| | | 211 | | { |
| | 0 | 212 | | arg += '\r'; |
| | 0 | 213 | | break; |
| | | 214 | | } |
| | | 215 | | case 't': |
| | | 216 | | { |
| | 0 | 217 | | arg += '\t'; |
| | 0 | 218 | | break; |
| | | 219 | | } |
| | | 220 | | case 'v': |
| | | 221 | | { |
| | 0 | 222 | | arg += '\v'; |
| | 0 | 223 | | break; |
| | | 224 | | } |
| | | 225 | | case '\\': |
| | | 226 | | { |
| | 1 | 227 | | arg += '\\'; |
| | 1 | 228 | | break; |
| | | 229 | | } |
| | | 230 | | case '\'': |
| | | 231 | | { |
| | 0 | 232 | | arg += '\''; |
| | 0 | 233 | | break; |
| | | 234 | | } |
| | | 235 | | case 'e': // Not ANSI-C, but used by bash. |
| | | 236 | | { |
| | 0 | 237 | | arg += '\u001B'; |
| | 0 | 238 | | break; |
| | | 239 | | } |
| | | 240 | | |
| | | 241 | | // |
| | | 242 | | // Process up to three octal digits. |
| | | 243 | | // |
| | | 244 | | case '0': |
| | | 245 | | case '1': |
| | | 246 | | case '2': |
| | | 247 | | case '3': |
| | | 248 | | case '4': |
| | | 249 | | case '5': |
| | | 250 | | case '6': |
| | | 251 | | case '7': |
| | | 252 | | { |
| | | 253 | | const string octalDigits = "01234567"; |
| | 1 | 254 | | short s = 0; |
| | | 255 | | int j; |
| | | 256 | | for ( |
| | 1 | 257 | | j = i; |
| | 1 | 258 | | j < i + 3 && |
| | 1 | 259 | | j < l.Length && |
| | 1 | 260 | | octalDigits.Contains(c = l[j], StringComparison.Ordinal); |
| | 1 | 261 | | ++j) |
| | | 262 | | { |
| | 1 | 263 | | s = (short)((s * 8) + c - '0'); |
| | | 264 | | } |
| | 1 | 265 | | i = j - 1; |
| | 1 | 266 | | arg += (char)s; |
| | 1 | 267 | | break; |
| | | 268 | | } |
| | | 269 | | |
| | | 270 | | // |
| | | 271 | | // Process up to two hex digits. |
| | | 272 | | // |
| | | 273 | | case 'x': |
| | | 274 | | { |
| | | 275 | | const string hexDigits = "0123456789abcdefABCDEF"; |
| | 1 | 276 | | if (i < l.Length - 1 && !hexDigits.Contains(l[i + 1], StringComparison.Ordinal)) |
| | | 277 | | { |
| | 0 | 278 | | arg += '\\'; |
| | 0 | 279 | | arg += 'x'; |
| | 0 | 280 | | break; |
| | | 281 | | } |
| | | 282 | | |
| | 1 | 283 | | short s = 0; |
| | | 284 | | int j; |
| | | 285 | | for ( |
| | 1 | 286 | | j = i + 1; |
| | 1 | 287 | | j < i + 3 && |
| | 1 | 288 | | j < l.Length && |
| | 1 | 289 | | hexDigits.Contains(c = l[j], StringComparison.Ordinal); |
| | 1 | 290 | | ++j) |
| | | 291 | | { |
| | 1 | 292 | | s *= 16; |
| | 1 | 293 | | if (char.IsDigit(c)) |
| | | 294 | | { |
| | 1 | 295 | | s += (short)(c - '0'); |
| | | 296 | | } |
| | 1 | 297 | | else if (char.IsLower(c)) |
| | | 298 | | { |
| | 0 | 299 | | s += (short)(c - 'a' + 10); |
| | | 300 | | } |
| | | 301 | | else |
| | | 302 | | { |
| | 1 | 303 | | s += (short)(c - 'A' + 10); |
| | | 304 | | } |
| | | 305 | | } |
| | 1 | 306 | | i = j - 1; |
| | 1 | 307 | | arg += (char)s; |
| | 1 | 308 | | break; |
| | | 309 | | } |
| | | 310 | | |
| | | 311 | | // |
| | | 312 | | // Process control-chars. |
| | | 313 | | // |
| | | 314 | | case 'c': |
| | | 315 | | { |
| | 1 | 316 | | if (i == l.Length - 1) |
| | | 317 | | { |
| | | 318 | | // Nothing after the \c: keep it unchanged. The missing closing quote is |
| | | 319 | | // reported after the loop. |
| | 1 | 320 | | arg += "\\c"; |
| | 1 | 321 | | break; |
| | | 322 | | } |
| | | 323 | | |
| | 1 | 324 | | c = l[++i]; |
| | 1 | 325 | | if ( |
| | 1 | 326 | | (char.ToUpper(c, CultureInfo.InvariantCulture) >= 'A' && |
| | 1 | 327 | | char.ToUpper(c, CultureInfo.InvariantCulture) <= 'Z') || |
| | 1 | 328 | | c == '@' || |
| | 1 | 329 | | (c >= '[' && c <= '_')) |
| | | 330 | | { |
| | 1 | 331 | | arg += (char)(char.ToUpper(c, CultureInfo.InvariantCulture) - '@'); |
| | | 332 | | } |
| | | 333 | | else |
| | | 334 | | { |
| | | 335 | | // |
| | | 336 | | // Bash does not define what should happen if a \c |
| | | 337 | | // is not followed by a recognized control character. |
| | | 338 | | // We simply treat this case like other unrecognized |
| | | 339 | | // escape sequences, that is, we preserve the escape |
| | | 340 | | // sequence unchanged. |
| | | 341 | | // |
| | 0 | 342 | | arg += '\\'; |
| | 0 | 343 | | arg += 'c'; |
| | 0 | 344 | | arg += c; |
| | | 345 | | } |
| | 0 | 346 | | break; |
| | | 347 | | } |
| | | 348 | | |
| | | 349 | | // |
| | | 350 | | // If inside an ANSI-quoted string, a backslash isn't followed by |
| | | 351 | | // one of the recognized characters, both the backslash and the |
| | | 352 | | // character are preserved. |
| | | 353 | | // |
| | | 354 | | default: |
| | | 355 | | { |
| | 1 | 356 | | arg += '\\'; |
| | 1 | 357 | | arg += c; |
| | 1 | 358 | | break; |
| | | 359 | | } |
| | | 360 | | } |
| | | 361 | | break; |
| | | 362 | | } |
| | | 363 | | case '\'': // End of ANSI-quote mode. |
| | | 364 | | { |
| | 1 | 365 | | state = State.Normal; |
| | 1 | 366 | | break; |
| | | 367 | | } |
| | | 368 | | default: |
| | | 369 | | { |
| | 1 | 370 | | arg += c; // Everything else is taken literally. |
| | | 371 | | break; |
| | | 372 | | } |
| | | 373 | | } |
| | | 374 | | break; |
| | | 375 | | } |
| | | 376 | | default: |
| | | 377 | | { |
| | | 378 | | Debug.Assert(false); |
| | | 379 | | break; |
| | | 380 | | } |
| | | 381 | | } |
| | | 382 | | } |
| | | 383 | | |
| | | 384 | | switch (state) |
| | | 385 | | { |
| | | 386 | | case State.Normal: |
| | | 387 | | { |
| | 1 | 388 | | vec.Add(arg); |
| | 1 | 389 | | break; |
| | | 390 | | } |
| | | 391 | | case State.SingleQuote: |
| | | 392 | | { |
| | 1 | 393 | | throw new ParseException("missing closing single quote"); |
| | | 394 | | } |
| | | 395 | | case State.DoubleQuote: |
| | | 396 | | { |
| | 1 | 397 | | throw new ParseException("missing closing double quote"); |
| | | 398 | | } |
| | | 399 | | case State.ANSIQuote: |
| | | 400 | | { |
| | 1 | 401 | | throw new ParseException("unterminated $' quote"); |
| | | 402 | | } |
| | | 403 | | default: |
| | | 404 | | { |
| | | 405 | | Debug.Assert(false); |
| | | 406 | | break; |
| | | 407 | | } |
| | | 408 | | } |
| | | 409 | | |
| | 1 | 410 | | return vec.ToArray(); |
| | | 411 | | } |
| | | 412 | | } |